UAV Video Coverage Quality Maps and Prioritized Indexing for Wilderness Search and Rescue Bryan S. Morse, Cameron H. Engh, and Michael A. Goodrich Department of Computer Science Brigham Young University Provo, Utah, United States Email: morse@byu.edu Abstract—Video-equipped mini unmanned aerial vehicles (mini-UAVs) are becoming increasingly popular for surveil- lance, remote sensing, law enforcement, and search and rescue operations, all of which rely on thorough coverage of a target observation area. However, coverage is not simply a matter of seeing the area (visibility) but of seeing it well enough to allow detection of targets of interest, a quality we here call “see-ability”. Video flashlights, mosaics, or other geospatial compositions of the video may help place the video in context and convey that an area was observed, but not necessarily how well or how often. This paper presents a method for using UAV-acquired video georegistered to terrain and aerial reference imagery to create geospatial video coverage quality maps and indices that indicate relative video quality based on detection factors such as image resolution, number of observations, and variety of viewing angles. When used for offline post-analysis of the video, or for online review, these maps also enable geospatial quality-filtered or prioritized non- sequential access to the video. We present examples of static and dynamic see-ability coverage maps in wilderness search- and-rescue scenarios, along with examples of prioritized non- sequential video access. We also present the results of a user study demonstrating the correlation between see-ability computation and human detection performance. Keywords-unmanned aerial vehicles, wilderness search and rescue, coverage quality maps, video indexing I. I NTRODUCTION Small lightweight mini-UAVs with 5–8 foot wingspans have seen increased use recently for aerial sensing due to their lower cost and ease of deployment. When equipped with a video camera and transmitter, these mini-UAVs can be used for surveillance, remote sensing, law enforcement, and search and rescue operations, all of which require rapid and thorough coverage of a target area. However, because of their lightweight nature, these aerial sensing platforms are highly unstable and easily buffeted by wind, and the operator’s intentions may not always correspond to the actual flight path. This makes it difficult for operators or video analysts to correctly determine what spatial areas were observed during a flight or sequence of multiple flights. In addition to covering the target area, it is also essential to maintain sufficient resolution to allow human operators to accomplish their task. Since the altitude and orientation of the plane are highly variable due to wind or other factors, so too is the resolution of the resulting video. As the plane banks to one side or the other, even an otherwise downward- pointing camera may end up seeing areas far away and at an oblique angle. This is compounded in varying terrain since the UAV’s height above ground may change rapidly even while maintaining constant altitude. One can try to maintain a consistent height above ground either manually or through automated means, but this is still subject to the limitations of the plane’s ability to climb or safely descend. Some flight paths, especially in difficult terrain, may make a low-altitude pass over the target area then maneuver to make another pass, providing only periodically usable video. Our work in this area has focused on using mini-UAVs to assist in Wilderness Search and Rescue (WiSAR) oper- ations [1]. Field trials [2] tell us that it is often difficult to tell what areas have been searched well. This assessment is an essential component of search-and-rescue applications because it is basically a prioritized search, focusing on the regions most likely to include the missing person. Also important to this task is the ability to efficiently review previously acquired video, perhaps in response to a search observation or during post hoc offline review. This can be made more efficient by providing users with the ability to intelligently access search video not only by georeferenced indexing but by coverage quality as well, allowing users to directly access usable observations of a specified target area. Assessing the usability and coverage of aerial video is a matter not only of whether the plane’s camera could see a point but how well it saw it. Once the video is georegistered to the underlying terrain, determining whether the camera saw specific points is a simple matter of viewing geometry, what we typically think of as “visibility”. But visibility- based coverage alone isn’t enough to determine how useful the video is—one must consider the viewing resolution as well as the number of times seen, the variation of viewing angle (which can often play a role in detection), etc. We call this latter quality “see-ability”. This paper presents a method for creating coverage quality maps based on see-ability that convey not only the video coverage of each part of a target area but also how useful that video information is for the person viewing it (Figure 1). Such coverage maps are useful for post hoc evaluation of the search, for planning either during or between flights, and for coordination with other team members. 978-1-4244-4893-7/10/$25.00 © 2010 IEEE 227